Multi-Perspective Fuzzy Reasoning and XGBoost-Based Analysis of Online Learning Behavior

Area of Science:

• Education Technology
• Artificial Intelligence in Education
• Learning Analytics

Background:

• Online education is rapidly expanding, necessitating effective methods to analyze student learning behavior.
• Understanding student engagement and performance is crucial for optimizing online teaching strategies and providing tailored support.

Purpose of the Study:

• To develop a comprehensive model for analyzing online student learning behavior from multiple perspectives.
• To accurately classify student emotions expressed in online learning environments.
• To facilitate early identification of students at risk of falling behind in online courses.

Main Methods:

• Data preprocessing from online teaching platforms.
• Construction of a multi-perspective fuzzy reasoning model (curriculum, individual, class dimensions).
• Development of an improved XGBoost algorithm (optimized via grey wolf optimization) for emotion classification.

Main Results:

• The fuzzy model effectively evaluated learning performance across different dimensions.
• The improved XGBoost algorithm achieved 98.78% maximum accuracy in emotion classification, outperforming comparison models.
• Analysis indicated significant student procrastination in task completion, with completion rates dropping sharply after deadlines.

Conclusions:

• The developed model provides effective analysis of online learning behavior.
• Early identification of at-risk students is possible, enabling personalized teaching and precise interventions.
• The research contributes to optimizing online education through data-driven insights into student behavior and emotions.